chapter 5.6: the network layer in the internet

what are the 10 principles that drove the design of the internet

from most to least important:


1. make sure it works
2. keep it simple
3. make clear choices
4. exploit modularity
5. expect heterogeneity
6. avoid static options and parameters
7. look for a good design, doesn’t need to be perfect
8. be strict when sending, tolerant when receiving
9. think about scalability
10. consider performance and cost

true or false: the internet can be viewed as an interconnected collection of CASes (Autonomous Systems) in the network layer

true: or a collection of networks

what is a Tier 1 network

networks used by every other network to reach the rest of the internet, the largest backbones of the internet structure

what is a backbone constructed of

high-bandwidth lines and fast routers attached to internet service providers (ISPs)

how does communication in the Internet work

* transport layer takes data streams and breaks them up so they may be sent as IP packets
* IP routers forward packet through the Internet until destination is reached
* at destination, the network layer reassembles all datagram parts and hands it to the transport layer, where errors may be corrected

theoretically, how large may IP packets be and why aren’t they this large

64 KB each

they’re usually no more than 1500 bytes so they can fit in one ethernet frame

true or false: IP provides error correction

false: it is a best effort service

true or false: IP provides reliability guarantees

false: it is a best effort service

what two main parts are there in an IPv4 datagram

1. header
2. payload or body (L4 transport layer protocol data unit)

how many bytes is the fixed part of IPv4 header

20 bytes

what is a ‘big-endian’ network byte order

bits are transmitted from left to right and top to bottom

what is the purpose of the version field in the IP header

tracks which version of the protocol the datagram belongs to, allows us to transition between different versions of the IP over time

true or false: IPv6 does not have a version field in its header

false

what is the purpose if the IHL field in IPv4

tells us how long the header is in 32 bit words (each bit represents a byte) since the header length is not constant

what is the minimum and maximum value of the IHL field in IPv4 and how many bytes does it represent

min: 5 bits, 20 bytes

max: 15 bits, 60 bytes

how long is the options field for IPv4

40 bytes

how many bits wide is the total length field in IPv4, what is the maximum value and what does that value represent

16 bits wide

max value: 65,535 bytes for total length of datagram including header

what is the purpose of the identification field in IPv4 header

allows destination host to determine which packet a newly arrived fragment belongs to, all fragments of a packet contain the same identification value

what is the purpose of the DF field in IPv4 header

DF = don’t fragment

tells a router whether or not to fragment the packet

what is the purpose of the MF field in IPv4 header

MF = more fragments

all fragments except the last one have this bit set so it is known whether all fragments have arrived

what is the purpose of the fragment offset field in IPv4 header

tells where in the current packet a fragment belongs

all fragments except last one must be a multiple of 8 bytes (the elementary fragment unit)

how many bits wide is the fragment offset field in IPv4, what is the maximum value and what does that value represent

13 bits wide

max: 8192 fragments per datagram

what is the purpose of the time to live field in IPv4 header, how many bits is it and what does its value represent

limits packet lifetime by counting down to 0, after which packet is discarded and warning packet returned to host

8 bits wide

originally value counted time in s but usually counts in hops

what is the purpose of the protocol field in IPv4 header, how many bits is it

tells which transport process to give the packet to such as connection-oriented TCP or connectionless UDP

it is 8 bits long

why is the time to live field necessary in the IPv4 header

routing tables may become corrupted, prevents packets lingering forever

state + describe some of the additional features that may be added in the options field for IPv4

* security: how secret the datagram is
* strict source routing: gives complete path to be followed
* loose source routing: gives list of routers not to be missed
* record route: makes each router append its IP address
* timestamp: makes each router append addresses and timestamp

how long are IPv4 addresses

32 bits

true or false: the IPv4 header checksum must be recomputed at each hop

true: since at least one field (time to live) changes hop-by-hop

true or false: IP addresses are hierarchical

true

true or false: ethernet addresses are hierarchical

false

what notation is used for IP addresses

dotted decimal notation

4x 8 bit numbers separated by dots

what are the two parts that make up an IP address and what do they represent

* network: what network the device is attached to
* host: the unique ID of the device within that network

what is a prefix

the part of the IP address that represents a contiguous block of IP address space corresponding to a network such as Ethernet LAN

how are prefixes written

the lowest IP address in the block followed by the size of the block

what determines the size of the prefix

the number of bits in the network portion of the IP address

where is prefix size determined in an IP address

it always follows an address

routing protocols carry prefixes to routers

what is a subnet mask

a binary mask of 1s corresponding to the prefix length for the network portion of the IP address

it is ANDed with the IP address to extract only the network portion (network address)

what is the subnet address

the part of IP address containing only the subnet

what is a key advantage of hierarchical IP addresses

they allow routers to forward packets based only on the network portion of the address, as long as each of the networks has a unique address block, hence reducing routing tables significantly

what are two disadvantages of hierarchical IP addresses

* host IP address depends on where it is located in the network, can’t be used anywhere in the world unlike Ethernet addresses
* wasteful of addresses unless carefully managed: too large network blocks will mean there are addresses that go to waste

what is subnetting and why is it used

splitting blocks of addresses into smaller groups of addresses for internal use as multiple networks

it allows for future expansion without being wasteful of addresses and without changing any external databases

what is the default-free zone

the part of the internet where core routers must know which way to go to get to every network

usually routers in ISPs and backbones in the middle of the internet

what is route aggregation and why is it used

combining multiple small prefixes into a single larger prefix

it is done to reduce routing table sizes

true or false: aggregation is an automatic process

true: it depends on which prefixes are located where in the internet and not on the administrator assigning addresses to networks

what is a supernet

opposite of subnet, as a result of route aggregation, creating a larger group of networks out of smaller groups

what is Classless InterDomain Routing (CIDR)

packets are sent in the direction of the most specific route or the longest matching prefix that has the fewest IP addresses

it helps reduce the size of the global

what is a variable length subnet masks

when the length of the prefix is varied to allow subnetting of the network whilst remaining in the boundaries of the network ID

what classful and special addressing

the old system of IP address formats that allowed a specific number of hosts and networks depending on the class specified

what are the two unique addresses for every subnet has

1. network address: host ID set to all 0s: 1 below host range
2. broadcast address: host ID set to all 1s: 1 above host range

what is the network address used for

commonly used for destination field in routing tables

what is the broadcast address used for

used to talk to all devices connected to the network at the same time

how many hosts do w have for if we have n host bits for address

2ⁿ-2

because we have 2 addresses reserved for the network and broadcast addresses

when does dynamic IP addressing not work

when devices are not switched off, e.g. servers, businesses etc

what is Network Address Translation (NAT) and why is it necessary

ISP assigns each home or business a single IP address for Internet traffic, within the network each host gets a unique IP address for intramural traffic, when a packet exits the customer network to be sent to the ISP to exit the customer network, an address translation to a shared public IP address takes place.

necessary because we are running out of IP addresses faster than we can implement IPv6

how does NAT work with TCP + UDP connections

TCP + UDP payloads have source and destination ports that indicates where the connection begins and ends, these ports are used to get the returning message back to the right place

what 7 things does problems does NAT introduce

1. violates architectural model of IP: not every machine uniquely identified
2. breaks end-to-end connectivity: incoming packets not accepted until after outgoing ones, unless NAT traversal techniques used
3. Internet no longer fully connectionless: maintaining connection state for each connection passing through is not connectionless
4. violates protocol layering rule: working at layer 3 looking into layer 4
5. since TCP/UDP not an internet requirement, NAT may fail
6. some applications use TCP/UDP in prescribed ways, meaning they will fail unless special precautions taken when setting up NAT
7. TCP source port field is 16 bits, and 4096 required for special uses, limiting number of NAT addresses used

how long are IPv6 addresses

128 bit / 16 bytes

what are the 9 major goals of IPv6

1. support billions of hosts, even with insufficient address allocation
2. reduce the size of the routing tables
3. simplify the protocol, to allow routers to process packets faster
4. provide better security (authentication and privacy)
5. pay more attention to the type of service, esp for real-time data
6. aid multicasting by allowing scopes to be specified
7. make it possible for a host to roam without changing its address
8. allow protocol to evolve in the future
9. permit the old and new protocol to coexist for years

how many header fields does IPv4 have

13

how many header fields does IPv6 have

7

what 5 major improvements does IPv6 have over IPv4

1. more addresses available
2. simplification of header allows faster processing of packets for better throughput and delay
3. better support for options, allow routers to skip options not intended for them, further speeding up packet process time
4. better security
5. better quality of service

true or false: a device that uses IPv4 can communicate with a device that speaks only IPv6

false

what are some examples of internet protocols

TCP, UDP, ICMP, IGMP, OSPF, BGP, DNS

true or false: IPv6 presents the options in a way which routers can skip over options not intended for them

true: this feature speeds up packet process time

true or false: IPv6 has major security benefits over IPv4

true

what is the purpose of the differentiated services field in IPv6

distinguishes the class of service for packets with different real-time delivery requirements

what are the lower two bits of the differentiated services field used for

signal explicit congestion indications

what is the purpose of the flow label in IPv6 header

provides a way for a source and destination to mark group of packets that have the same requirements + should be treated in the same way by the network to form a pseudoconnection

what is the benefit of the flow label in the IPv6 header

it provides the flexibility of a datagram network and the guarantees of a virtual-circuit network

what is the purpose of the payload length in IPv6 header

how many bytes follow the 40-byte header in the payload

what is the difference between the two length fields in IPv4 and IPv6

IPv4: total length

IPv6: payload length

what is the maximum payload length following the header for:

* IPv4
* IPv6

IPv4: 65,535 bytes

IPv6: 65,515 bytes

how big is the IPv6 header (no extensions)

40 bytes

what is the purpose of the next header field in IPv6 header

tells which of the six extension headers (if any) follow it

if it is the last one in IP header, it tells which transport handler (e.g. TCP or UDP) to pass the packet to

what is the purpose of the hop limit field in IPv6 header

similar to time to live field in IPv4

stops packets from living forever by assigning a lifetime to them

how are IPv6 addresses written

8 groups of 4 hexadecimal digits separated by colons

what does it represent when there are a pair of colons in an IPv6 address

at least one group of 0s

true or false: IPv6 has fields that relate to fragmentation

false

what fields deal with fragmentation and reassembly in IPv4 header

* identification
* DF (don’t fragment)
* MF (more fragments)
* fragment offset

which fields from the IPv4 header have been removed in IPv6

* protocol field (replaced by next header)
* all fields relating to fragmentation
* checksum

why has IPv6 removed all fields relating to fragmentation

if a message is too large for a router to forward, an error message is returned to host who will deal with the future fragmentation of a message

more efficient

what are the six options for extension headers in IPv6

hop-by-hop options: misc. info for routers along path

destination options: additional info for destination

routing: loose list of routers to visit

fragmentation: management of datagram fragments

authentication: verification of sender’s identity

encrypted security payload: info about encrypted contents

how are IPv6 header options with variable-length encoded

(type, length, value) tuple

what is the purpose of the encoded type field in the variable-length header options for IPv6

1-byte field showing which option of extension it is

last 2 bits tell routers what to do with it if they do not know how to process the option (skip, discard packet, discard and send back ICMP)

what happens to the payload length field when the hop-by-hop header option is used in IPv6

it is set to 0

true or false: a next header field is added before every optional extension header in IPv6

true

what does the addition of the routing header do in IPv6 and how is it different to IPv4 loose source routing

it lists one or more routers in order that must be visited on the packet’s trip to the destination

unlike IPv4, other routers not listed may be visited in between

true or false: only hosts can fragment a message in IPv6

true

what are dual-stack hosts

hosts that have an IPv4 and IPv6 implementation so they can select a protocol to use depending on packet destination